Details

Autor(en) / Beteiligte

• Rahman, Shanto
• Li, Chengpeng
• Shi, August

Titel

TSVD4J: Thread-Safety Violation Detection for Java

Ist Teil von

• 2023 IEEE/ACM 45th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion), 2023, p.78-82

Ort / Verlag

Piscataway, NJ, USA: IEEE Press

Erscheinungsjahr

2023

Link zum Volltext

• IEEE_Xplore

Quelle

IEEE Electronic Library (IEL)

Beschreibungen/Notizen

• Concurrency bugs are difficult to detect and debug. One class of concurrency bugs are thread-safety violations, where multiple threads access thread-unsafe data structure at the same time, resulting in unexpected behavior. Prior work proposed an approach TSVD to detect thread-safety violations. TSVD injects delays at API calls that read/write to specific thread-unsafe data structures, tracking whether multiple threads can overlap in their accesses to the same data structure through the delays, showing potential thread-safety violations. We additionally enhance the TSVD approach to also consider read/write operations to object fields. We implement the TSVD approach in Java in our tool TSVD4J. TSVD4J can be integrated as a Maven plugin that can be included in any Maven-based application. Our evaluation on 12 applications shows that TSVD4J can detect 55 pairs of code locations accessing the same shared data structure across multiple threads, representing potential thread-safety violations. We find that the addition of tracking field accesses contributed the most to detecting these pairs. TSVD4J also detects more such pairs than existing tool RV-Predict. The demo video for TSVD4J is available at https://www.youtube.com/watch?v=-wSMzlj5cMY.